本文对旋转机械振动故障预

测方法进行了初步研究，分析了传统预测参数的不足，提出了以分时尺度矩为预测故

障参数的两种故障预测模型:分时尺度矩趋势预测模型和分时尺度矩波动预测模型。

根据分时尺度矩的变化不仅能够判断是否发生了故障，而且能够判断故障的种类。典

型故障实测数据的预测结果验证了基于分时尺度矩预测参数的灰色模型故障预测算

法的有效性。

-This paper the grey correlation analysis method and the probabilistic neural network

The shaft of rotating machinery typical fault the fusion diagnosis research. This paper puts forward the concept of correction coefficient of correlation,

The traditional grey correlation algorithm was improved, and achieved good diagnosis. In addition, this paper points scale